Towards a Landmark-based Flat Routing Rafael Pasquini 1 ,F´ abio L. Verdi 1 and Maur´ ıcio F. Magalh˜ aes 1 1 Department of Computer Engineering and Industrial Automation (DCA) School of Electrical and Computer Engineering (FEEC) State University of Campinas (Unicamp) C. P. 6.101 – 13.083-970 – Campinas – SP – Brazil {pasquini,verdi,mauricio}@dca.fee.unicamp.br Abstract. Two main groups of Flat Routing proposals are found in the literature. The biggest group considers the existence of an underlay network providing di- rect communication between neighbors at the flat identity layer. On the other hand, a smaller set of proposals consider a scenario that has none underlay net- work, i.e., routing directly on flat identifiers/names. Our interest is concentrated on the second group due to the perspective of a new internetworking model in which the network layer has no information regarding location and, in this pa- per, our Landmark-based Flat Routing proposal is introduced. We also present a tool for evaluating different topologies and flat routing protocols. The quan- titative results show the signaling overhead and the trade-off between routing table size and route stretch. The results were collected using the tool under two distinct topologies, a regular mesh and an Internet-like topology. 1. Introduction The investigative scenario of this paper is related to routing directly on flat identi- fiers/names. This approach considers that the network layer does not contain any lo- cation information to drive the packet forwarding. Examples of proposals in this sce- nario include Identity Based Routing (IBR) [Caesar 2007], Virtual Ring Routing (VRR) [Caesar et al. 2006a] and Routing on Flat Labels (ROFL) [Caesar et al. 2006b]. Some of the benefits introduced by this approach include: 1) native support for mobility and multi- homing; 2) simpler allocation of identifiers by requiring only uniqueness (the IP requires uniqueness and topological adherence); 3) no need for new mapping services since data is forwarded based on the flat identity (no need for identifier to locator mapping) and 4) better support for network access controls which can be applied on the identifier. In this line, our Landmark-based Flat Routing (LFR) proposal is introduced. The landmark term comes from Compact Routing 1 [Krioukov et al. 2007] work and its defini- tion found in [Tsuchiya 1988] is: “A Landmark is a router whose neighbor routers within a certain number of hops contain routing entries for that router”. In order to investigate the features and the behavior of our proposal, we developed a tool that is capable of emulating flat routing without the usage of any address-based substrate. The ideas behind this tool are: 1) development of flat routing mechanisms for verifying its behaviors under different scenarios and 2) simplification of the analysis for 1 Refer to the Dmitri work [Krioukov et al. 2007] as a starting point for the vast amount of Compact Routing information available in the literature.